Method and apparatus for enhancing 3D model resolution

Abstract

Systems and methods of enhancing the resolution or restoring details associated with high resolution images into a filtered digital surface model (DSM) for location-based applications and analyses. The disclosed methods include mapping the changing gray scale values (intensity) from the images to changes in elevation in the DSM using a regression over a local neighborhood of pixels. Further, the disclosed methods do not rely on information about the sensor illumination geometry, and are extendable to be able to utilize any types of images. Additionally, the disclosed embodiments are sensor agnostic. That is, the disclosed methods can be applied on any type of images collected by any type of sensor.

The invention claimed is: 
     
       1. A computer-implemented method comprising:
 receiving a first model associated with an original image; 
 resampling the first model for one or more pixels in the original image; 
 creating a difference model based on subtracting a low frequency model from the resampled first model; 
 generating a map based on correlating at least one pixel in the difference model to a group of pixels in the original image; and 
 constructing a second model by manipulating grayscale values of the one or more pixels in the original image based on the map. 
 
     
     
       2. The method of  claim 1 , wherein resampling the first model is based at least on: bilinear resampling, bicubic resampling, nearest neighbor resampling, natural neighbor resampling, kriging resampling, box average resampling, or box median resampling. 
     
     
       3. The method of  claim 1 , wherein resampling the first model includes determining that each pixel in the resampled first model is associated with a corresponding pixel in the original image. 
     
     
       4. The method of  claim 1 , further comprising:
 upon determining that the original image is not in a grayscale format, converting the original image to a grayscale format for extracting the grayscale values of the one or more pixels in the original image. 
 
     
     
       5. The method of  claim 1 , wherein the low frequency model is generated using at least one of: a Digital Terrain Model (DTM), a Fourier Domain filter that removes the high frequency content, a spatial smoothing filter. 
     
     
       6. The method of  claim 1 , wherein the low frequency model is generated using a smoothing filter, wherein the parameters of the smoothing filter (i) prevent over-smoothing features included in the original model and (ii) remove one or more surface features from the original model. 
     
     
       7. The method of  claim 1 , wherein the low frequency model is generated using a Fourier domain filter that isolates low-order variations from the first model. 
     
     
       8. The method of  claim 1 , wherein the neighborhood adjacent to each pixel in the difference model is a kernel of a predetermined size. 
     
     
       9. The method of  claim 1 , wherein the resolution of the second model is higher than a resolution of the first model. 
     
     
       10. The method of  claim 1 , wherein the resolution of the second model is invariant of slopes of features included in the original image, and wherein a resolution of the second model is invariant to scale changes in the original image or characteristics of sensors associated with capturing the original image. 
     
     
       11. The method of  claim 10 , wherein the second model is constructed contemporaneously to receiving the first model. 
     
     
       12. The method of  claim 1 , wherein the original image is at least one of: a thermal image, a multi-spectral image, a hyper-spectral image, an optical image, a medical image, a radar image, a weather image, a fused image from multiple types of sensors, a color image, a gray scale image, or a LiDAR intensity image. 
     
     
       13. The method of  claim 1 , wherein the first model and the second model correspond to a digital surface model (DSM), a tomographic model, a weather model, or a three dimensional (3D) digital model. 
     
     
       14. The method of  claim 1 , wherein manipulating the grayscale values based on the correspondence table includes generating a regression table from the correspondence table. 
     
     
       15. The method of  claim 14 , wherein the regression model corresponds to a linear model. 
     
     
       16. The method of  claim 14 , wherein the regression model corresponds a non-linear model. 
     
     
       17. A non-transitory computer-readable storage medium storing instructions configured to cause at least one computing device to perform a method comprising:
 receiving a first model associated with an original image; 
 resampling the first model for one or more pixels in the original image; 
 creating a difference model based on subtracting a low frequency model from the resam pled first model; 
 generating a map based on correlating at least one pixel in the difference model to a group of pixels in the original image; and 
 constructing a second model by manipulating grayscale values of the one or more pixels in the original image based on the map. 
 
     
     
       18. The computer-readable storage medium of  claim 17 , wherein the original image is at least one of: a thermal image, a multi-spectral image, a hyper-spectral image, an optical image, a medical image, a radar image, a weather image, a fused image from multiple types of sensors, a color image, a gray scale image, or a LiDAR intensity image. 
     
     
       19. A computer system comprising:
 at least one processor and 
 at least one memory comprising instructions configured to cause the at least one processor to perform a method comprising:
 receiving a first model associated with an original image; 
 resampling the first model for one or more pixels in the original image; 
 creating a difference model based on subtracting a low frequency model from the resampled first model; 
 generating a map based on correlating at least one pixel in the difference model to a group of pixels in the original image; and 
 constructing a second model by manipulating grayscale values of the one or more pixels in the original image based on the map. 
 
 
     
     
       20. The system of  claim 19 , wherein the low frequency model is generated using at least one of: a Digital Terrain Model (DTM), a Fourier Domain filter that removes the high frequency content, or a spatial smoothing filter.